RIBOTAC-mediated degradation of hsa-microRNA-301a-3p suppresses TNBC bone metastasis

Abstract

Triple-negative breast cancer (TNBC) is an aggressive subtype with a high propensity for bone metastasis and limited treatment options. Although many proteins drive metastasis, most remain inaccessible to conventional small-molecule drugs due to lack of suitable binding sites, prompting interest in upstream RNA regulators. Using integrative bioinformatics and machine learning, we identified hsa-microRNA-301a-3p as a central regulator of TNBC bone metastasis. However, current nucleic acid–based therapies are limited by instability and inefficient delivery. To overcome these challenges, we designed a ribonuclease-targeting chimera (RIBOTAC) that recruits endogenous RNase L to degrade the primary transcript pri-microRNA-301a. The resulting RIBOTAC efficiently reduced hsa-microRNA-301a-3p levels, restored NEDD4L expression, and promoted degradation of the oncogenic protein YAP, thereby suppressing TNBC bone metastasis. These findings identify hsa-microRNA-301a-3p as a key driver and demonstrate RNA-targeted degradation as a therapeutic strategy in TNBC bone metastasis.

INTRODUCTION

Triple-negative breast cancer (TNBC) represents one of the most aggressive breast cancer subtypes, accounting for 15–20% of all cases and characterized by early recurrence, visceral and bone metastases, and limited responsiveness to current systemic therapies (1–5). Bone is among the most common and debilitating metastatic sites, where lesions cause fractures, hypercalcemia, and severe pain, severely compromising patients’ quality of life (3, 5). Despite advances in chemotherapy, immunotherapy, and antibody–drug conjugates, effective strategies that specifically prevent or treat TNBC bone metastasis remain lacking.

Emerging evidence indicates that dysregulation of microRNAs (miRNAs) plays a pivotal role in the initiation and progression of breast cancer (6–8). By controlling post-transcriptional gene expression, miRNAs shape key processes including cell proliferation, migration, epithelial–mesenchymal transition (EMT), and metastatic colonization (6). Several oncogenic miRNAs—such as hsa-microRNA-10b, hsa-microRNA-21, and hsa-microRNA-155—have been implicated in breast cancer metastasis (9, 10). However, the specific miRNA circuits driving TNBC progression and bone colonization are still not well defined. Clarifying these TNBC-specific oncogenic miRNA networks may reveal critical determinants of metastasis and therapeutic vulnerabilities (8).

Among candidate oncogenic miRNAs, hsa-microRNA-301a-3p has emerged as a potent driver of tumor aggressiveness across multiple malignancies, including breast, pancreatic, renal, gastric, and lung malignancies (6, 11–16). Elevated hsa-microRNA-301a-3p expression correlates with advanced tumor stage and poor overall survival in breast cancer patients (14–16). Mechanistically, hsa-microRNA-301a-3p targets multiple tumor-suppressive genes, such as PTEN, SMAD4, RUNX3, INHBA, BTG1, and PHD3, thereby activating various downstream signaling pathways that support cancer progression (3, 12, 13, 17–20). Although bioinformatic analyses suggested regulatory circuits involving hsa-microRNA-301a-3p in breast cancer, these predictions remain unvalidated, and its role in TNBC bone metastasis has not been systematically investigated (21). Despite its therapeutic potential, the non-coding nature of hsa-microRNA-301a-3p hinders conventional drug development, as it lacks the well-defined binding pockets required for traditional small-molecule targeting. This limitation highlights a broader challenge in designing pharmacological strategies against pathogenic RNAs.

Recent advances in selective RNA degradation have enabled pharmacological strategies for directly eliminating oncogenic RNAs in cancer (22). Among these approaches, ribonuclease-targeting chimeras (RIBOTACs) have emerged as an innovative platform for small-molecule-driven RNA degradation (22–26). RIBOTACs are bifunctional molecules that couple an RNA-binding ligand to an RNase L recruiter, enabling selective cleavage of target RNAs by leveraging endogenous ribonuclease activity (23, 26). Compared with antisense oligonucleotides or siRNAs, RIBOTACs broaden the druggable transcriptome to include otherwise “undruggable” targets such as oncogenic transcription factors and non-coding RNAs. They also exhibit substantially greater selectivity and potency than their parent ligands (e.g., >2,500-fold for pre-microRNA-21 and >10-fold for C5-RIBOTAC), and demonstrate robust efficacy across multiple disease models (26–31). Although several RIBOTACs have been developed to degrade pathogenic RNAs, their application to solid tumors, particularly for targeting oncogenic miRNAs, remains largely unexplored and represents a promising direction for therapeutic development.

In this study, structure-guided modeling and molecular docking revealed that Amikacin binds the Drosha cleavage site of pri-microRNA-301a, enabling the design of an Amikacin-based RIBOTAC that recruits RNase L to degrade the target RNA. In cellular and in-vivo models of TNBC bone metastasis, this RIBOTAC markedly reduced hsa-microRNA-301a-3p expression, restored the tumor-suppressive E3 ubiquitin ligase NEDD4L, and promoted proteasomal degradation of YAP1, thereby robustly suppressing metastatic dissemination. Collectively, our findings establish hsa-microRNA-301a-3p as a key oncogenic driver of TNBC metastasis and provide a proof-of-concept for its selective degradation using small molecules. This study highlights the therapeutic potential of RIBOTACs as a pharmacological strategy for targeting oncogenic non-coding RNAs in aggressive and metastatic breast cancers.

RESULTS

Identification of hsa-microRNA-301a-3p as a key oncogenic microRNA driving bone metastasis in TNBC

To identify microRNAs associated with TNBC malignancy, we first performed integrative analysis of miRNA expression profiles from our in-house RNA sequencing datasets and The Cancer Genome Atlas (TCGA) cohort. Specifically, upregulated microRNAs were identified by comparing MDA-MB-231 or MDA-BoM-1833 cells with MCF10A, as well as TNBC tumor versus normal tissues in the TCGA cohort. Intersection of these datasets yielded six consistently upregulated microRNAs (Fig. 1A). Univariate Cox regression analysis of these six candidate microRNAs revealed that hsa-microRNA-301a-3p displayed the most significant association with poor prognosis (Fig. 1B). Additional pan-cancer expression and survival analyses were provided in fig. S1. Machine learning–based feature selection using least absolute shrinkage and selection operator (LASSO), random survival forest (RSF), and gradient boosting model (GBM) based on these candidates consistently ranked hsa-microRNA-301a-3p as the top risk factor (Fig. 1C). To investigate its functional relevance, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of predicted target genes of hsa-microRNA-301a-3p. These genes were significantly enriched in pathways implicated in metastasis, including Hippo, PI3K-Akt, and TGF-β signaling (Fig. 1D). Functional enrichment analysis further demonstrated that these genes were predominantly associated with metastasis, proliferation, immune regulation, and apoptosis, with metastasis-related processes showing the highest enrichment (Fig. 1E). Consistent with these findings, hsa-microRNA-301a-3p expression correlated with metastatic phenotypes across multiple organs, with the strongest association observed in bone metastasis, supporting its role as a key driver of TNBC bone metastasis (Fig. 1F). Quantitative reverse transcription PCR (qRT-PCR) analysis across a panel of breast cell lines showed that hsa-microRNA-301a-3p expression progressively increased from normal mammary epithelial MCF10A cells to parental MDA-MB-231 cells and reached the highest level in bone-metastatic MDA-BoM-1833 cells (Fig. 1G).

Fig. 1. Functional validation of hsa-microRNA-301a-3p in TNBC.

(A) Venn diagram showing the overlap of upregulated microRNAs from three datasets: MDA-MB-231 vs MCF10A (gray), MDA-BoM-1833 vs MCF10A (purple), and TCGA-TNBC cohort (red). Six commonly upregulated microRNAs were identified. (B) Univariate Cox regression analysis of the six candidate microRNAs. Forest plot showing hazard ratios (HRs) and 95% confidence intervals, identifying hsa-microRNA-301a-3p as significantly associated with poor prognosis. (C) Feature importance ranking of candidate microRNAs using machine learning algorithms, including LASSO, RSF, and GBM. hsa-microRNA-301a-3p was consistently ranked as the top prognostic factor. (D) KEGG pathway enrichment analysis of predicted target genes of hsa-microRNA-301a-3p, highlighting significant enrichment in metastasis-related signaling pathways, including Hippo, PI3K-Akt, and TGF-β signaling. (E) Functional enrichment analysis of hsa-microRNA-301a-3p-targeted genes, showing predominant association with metastasis-related biological processes. (F) Correlation analysis between hsa-microRNA-301a-3p and organ-specific metastasis, indicating a strong association with bone metastasis. (G) Relative expression levels of hsa-microRNA-301a-3p in MCF10A, MDA-MB-231, and MDA-BoM-1833 cells. (H and I) Validation of knockdown efficiency of hsa-microRNA-301a-3p using inhibitor and shRNA approaches. (J) Representative images of wound healing and transwell migration assays following hsa-microRNA-301a-3p inhibition. (K and L) Quantification of migration corresponding to (J). (M) Schematic illustration of the in vivo bone metastasis model established by intracardiac injection of MDA-BoM-1833 cells with stable knockdown of hsa-microRNA-301a-3p. Created in BioRender. Wei, Z. (2026) https://BioRender.com/17ydoyn. (N) Bioluminescence imaging of metastatic burden at indicated time points. (O) Quantification of bioluminescence signals at day 28. (P) TRAP staining and micro-CT imaging. (Q and R) ELISA analysis of bone metabolism markers, including BSP and β-CTX. Data are presented as mean ± SD. Statistical significance was assessed using Student’s t test in (G) to (I), (K) to (L), (O), and (Q) to (R) where *P < 0.05, **P < 0.01, ***P < 0.001.

Functional inhibition of hsa-microRNA-301a-3p using a specific microRNA inhibitor (table S1) was validated by qRT-PCR (Fig. 1H). Stable knockdown efficiency in MDA-BoM-1833 cells was further confirmed by qRT-PCR (Fig. 1I). These suppressive interventions markedly reduced cell migration in wound-healing and transwell assays (Fig. 1, J to L). In vivo, luciferase-labeled MDA-BoM-1833 cells with stable hsa-microRNA-301a-3p knockdown were injected into the left ventricle of nude mice to establish a bone-metastasis model (Fig. 1M). Bioluminescence imaging revealed a dramatic reduction of metastatic signals in the hsa-microRNA-301a-3p-silenced group (Fig. 1, N and O). Consistently, TRAP staining, micro-CT analyses and ELISA analysis demonstrated decreased osteolytic lesions and altered bone-resorption marker β-CTX and bone-turnover-associated marker BSP (Fig. 1, P to R).

Structure-guided design of an Amikacin-based RIBOTAC targeting pri-microRNA-301a in TNBC

Analysis of the secondary structure of pri-microRNA-301a revealed exposed loop regions near the Drosha cleavage site, suggesting the presence of defined structural pockets that could be targeted by small molecules. Based on this observation, we developed a structure-guided screening strategy to identify compounds capable of binding this region and enabling targeted RNA degradation (Fig. 2A). The 3D structure of pri-microRNA-301a was modeled using RNA structure and MC-Fold/MC-Sym platform (Fig. 2B), followed by molecular docking against multiple virtual libraries (Inforna, Coconut, and MOE) (32–33). This workflow identified 66 candidate compounds predicted to bind the Drosha cleavage region, among which Amikacin exhibited the most favorable binding free energy (ΔG ≈ −8 kcal/mol) and engaged both loop motifs flanking the cleavage site (Fig. 2C). Specifically, Amikacin established extensive intermolecular interactions with the yellow-labeled U-rich loop, engaging nucleotides A-74, C-76, and G-10. Meanwhile, it formed stable contacts with the green-labeled purine-rich loop, recognizing key residues including A-63, U-65, and A-20. Together, these interactions stabilized Amikacin within the structural pocket surrounding the Drosha cleavage site.

Fig. 2. Structure-guided design and biophysical validation of an Amikacin-based RIBOTAC targeting pri-microRNA-301a.

(A) Schematic workflow integrating RNA structural modeling and molecular docking to identify candidate small molecules binding to the Drosha cleavage site of pri-microRNA-301a. (B) Predicted 2D and 3D structure of pri-microRNA-301a. (C) Docking model of Amikacin bound to pri-microRNA-301a, showing an estimated binding free energy of approximately −8 kcal/mol at loop regions. (D) Synthetic route of RIBOTAC. (E) Schematic illustration of the RIBOTAC mechanism, consisting of an RNA binder, a flexible linker, and an RNase L-recruiting module. (F) Design of the pri-microRNA-301a RNA constructs derived from the C1 template by either replacing the Drosha cleavage region or introducing complementary mutations to form a closed loop. (G) MST analysis of the pri-microRNA-301a RNAs in (F) with RIBOTAC or Amikacin. (H and I) qRT-PCR analysis of hsa-microRNA-301a-3p levels in MDA-MB-231 cells (H) and MDA-BoM-1833 cells (I) following treatment with Amikacin or RIBOTAC at the indicated concentrations. Data are presented as mean ± SD (n = 3). Statistical significance was assessed using one-way ANOVA with Tukey’s multiple comparisons test in (H) to (I), where *P < 0.05, **P < 0.01, ***P < 0.001.

We subsequently designed and synthesized an Amikacin-based RIBOTAC by connecting Amikacin to an RNase L recruiter through a flexible linker (Fig. 2D). The detailed synthetic route was shown in fig. S2. Compound purity and structural characterization, confirmed by HPLC and MS, were provided in fig. S3. Mechanistically, the RIBOTAC is designed to recruit RNase L to the RNA substrate, promoting RNase L dimerization and inducing target RNA cleavage (Fig. 2E). To validate target engagement, we generated a series of RNA probes based on the C1 template, in which the Drosha cleavage region was either substituted or mutated to enforce base pairing and close the loop structure (Fig. 2F). All RNA probes were labeled at the 5′ end with Cy5 and analyzed by microscale thermophoresis (MST). MST measurements revealed that both Amikacin and the corresponding RIBOTAC bound with comparable affinities to the loop regions surrounding the Drosha cleavage site. In contrast, these interactions were completely abolished when the cleavage-loop motif was disrupted. The RNase L-recruiting moiety alone showed no detectable affinity with pri-microRNA-301a (Fig. 2G), confirming that RNA recognition was driven by the Amikacin scaffold. In line with this target-engagement model, RIBOTAC showed minimal anti-migratory activity in low-target-expressing MCF10A cells, whereas excess Amikacin attenuated the anti-migratory effect of RIBOTAC in MDA-BoM-1833 cells (fig. S4). Consistent with this mechanism, qRT-PCR analysis revealed that RIBOTAC treatment resulted in a significant, concentration-dependent reduction of mature hsa-microRNA-301a-3p levels in TNBC cells (Fig. 2, H and I).

RIBOTAC selectively degrades hsa-microRNA-301a-3p and suppresses TNBC cell migration in vitro

To confirm the RNA degradation activity of RIBOTAC, we established a fluorescence-based RNase L cleavage assay using the RNA substrate labeled with a fluorophore (FAM) at the 5′ end and a quencher (IQ4) at the 3′ end. In this system, fluorescence remained quenched until RNA was cleaved (Fig. 3A and fig. S5). RIBOTAC treatment produced a dose-dependent increase in fluorescence, which was higher than with Amikacin or the RNase L recruiter alone (Fig. 3B). Consistent with these findings, flow cytometric analysis showed a pronounced increase in intracellular FAM fluorescence in both MDA-MB-231 and MDA-BoM-1833 cells following RIBOTAC treatment, indicating enhanced intracellular RNA cleavage (Fig. 3, C to F). Functionally, RIBOTAC treatment markedly inhibited cell migration (Fig. 3, G and H) and suppressed clonogenic growth in a dose-dependent manner (fig. S6). Real-time impedance-based monitoring further demonstrated a dose-dependent reduction in the cell motility index (Fig. 3, I and J). Western blot analysis showed dose-dependent RNase L dimerization induced by RIBOTAC (Fig. 3, K and L). RIBOTAC treatment led to the most significant downregulation of hsa-microRNA-301a-3p in microRNA sequencing, with the smallest P-value and high reproducibility (Fig. 3M, fig. S7A). The global miRNA heatmap further showed that most detectable non-target miRNAs were not broadly depleted, with only a limited subset showing moderate expression changes (fig. S7A). Pre-inhibition of hsa-microRNA-301a-3p significantly attenuated the inhibitory effect of RIBOTAC on cell migration (Fig. 3, N to P). Similarly, a colony formation rescue assay further confirmed the dependence of RIBOTAC activity on hsa-microRNA-301a-3p (fig. S7, B and C). To further confirm RNase L dependency, we performed RNase L knockdown and rescue experiments in MDA-BoM-1833 cells. RNase L knockdown attenuated RIBOTAC-mediated suppression of hsa-microRNA-301a-3p and downstream NEDD4L/YAP1 regulation, whereas RNase L re-expression restored RIBOTAC responsiveness (fig. S8). Collectively, these results support RNase L-dependent target RNA suppression by RIBOTAC, leading to reduced hsa-microRNA-301a-3p levels and inhibited cell migration.

Fig. 3. RIBOTAC promotes hsa-microRNA-301a-3p degradation and suppresses cell migration.

(A) Schematic illustration of the fluorescence-based RNase L cleavage assay. (B) RIBOTAC treatment introduces a dose-dependent increase in fluorescence compared with Amikacin or RNase L recruiter alone. (C) Flow-cytometric analysis of FAM fluorescence in MDA-MB-231 cells treated with Amikacin or RIBOTAC. (D) Analysis of (C). (E) Flow-cytometric analysis of FAM fluorescence in MDA-BoM-1833 cells treated with Amikacin or RIBOTAC. (F) Analysis of (E). (G) Transwell assay of TNBC cells treated with Amikacin or RIBOTAC. (H) Quantification of percentage of migrated cells from (G). (I and J) RTCA assay to monitor motility of MDA-MB-231 (I) and MDA-BoM-1833 (J) cells. (K and L) Western blot analysis of RNase L dimerization in MDA-MB-231 (K) and MDA-BoM-1833 (L) cells following RIBOTAC treatment. (M) Volcano plot of miRNA-seq analysis showing differentially expressed miRNAs in MDA-MB-231 cells treated with RIBOTAC compared with control. (N) Representative images of transwell assays in MDA-MB-231 and MDA-BoM-1833 cells treated with RIBOTAC, NC inhibitor, or hsa-microRNA-301a-3p inhibitor. (O) Quantification of migrated MDA-MB-231 cells shown in (N). (P) Quantification of migrated MDA-BoM-1833 cells shown in (N). Data are presented as mean ± SD; n is indicated in the figure. Statistical significance was assessed using Student’s t test in (D) and (F) and one-way ANOVA with Tukey’s multiple comparisons test in (B), (H), (O), and (P), where *P < 0.05, **P < 0.01, ***P < 0.001.

RIBOTAC inhibits TNBC bone metastasis in vivo

We next evaluated the anti-metastatic efficacy of RIBOTAC in a bone-metastatic mouse model established by intracardiac (left ventricle) injection of luciferase-labeled MDA-BoM-1833 or 4T1.2 cells (Fig. 4). Mice were treated intravenously with RIBOTAC, paclitaxel or zoledronic acid (Fig. 4, A and I). No significant changes in body weight were observed during RIBOTAC treatment, and hemolysis assays demonstrated negligible red blood cell lysis (fig. S9), indicating good systemic tolerability (Fig. 4, B and J). Consistently, serial bioluminescence imaging confirmed a progressive and time-dependent suppression of metastatic burden following RIBOTAC treatment (Fig. 4, C, D, K, and L). Quantitative analysis showed a significant reduction in metastatic signal intensity and serum bone turnover-associated marker BSP and bone-resorption marker β-CTX in RIBOTAC-treated mice (Fig. 4, E, F, N, and O). Tartrate-resistant acid phosphatase (TRAP) staining and micro-CT imaging revealed a marked reduction in osteolytic lesions in RIBOTAC-treated groups compared to controls (Fig. 4, G, H, and M). Fluorescence in situ hybridization (FISH) and hematoxylin and eosin (HE) staining further confirmed reduced hsa-microRNA-301a-3p expression and diminished tumor burden in bone lesions, respectively (fig. S10).

Fig. 4. RIBOTAC reduces osteolytic bone lesions and tumor burden in TNBC bone metastasis models.

(A) Schematic illustration of the MDA-BoM-1833 TNBC bone metastasis model established via left ventricle injection. Created in BioRender. Wei, Z. (2026) https://BioRender.com/17ydoyn. (B) Body weight monitoring of mice. (C) Quantification of bioluminescent signal intensity at day 28. (D) In vivo bioluminescence imaging at the indicated time points. (E and F) ELISA detection of serum bone turnover-associated marker BSP (E) and bone-resorption marker β-CTX (F). (G) Representative TRAP staining and micro-CT reconstruction of bones. (H) Quantitative analysis of TRAP-positive area from (G). (I) Schematic illustration of the 4T1.2 TNBC bone metastasis model established via left ventricle injection. Created in BioRender. Wei, Z. (2026) https://BioRender.com/17ydoyn. (J) Body weight monitoring of mice. (K) Quantification of bioluminescent signal intensity at day 21. (L) In vivo bioluminescence imaging at indicated time points. (M) Representative micro-CT reconstruction of bones. (N and O) ELISA detection of serum bone turnover-associated marker BSP and bone-resorption marker β-CTX. Data are presented as mean ± SD; n is indicated in the figure. Statistical significance was assessed using Student’s t test in (C), (E), (F), (H), (K), (N), and (O) and one-way ANOVA with Tukey’s multiple comparisons test in (B) and (J), where *P < 0.05, **P < 0.01, ***P < 0.001.

RIBOTAC targets Hippo signaling to promote YAP1 degradation in TNBC

To systematically characterize the signaling pathways affected by RIBOTAC, we performed KEGG enrichment analyses based on downregulated genes identified from both mRNA-seq and miRNA-seq datasets following RIBOTAC treatment. Intersection analysis of the top-ranked pathways revealed that Hippo signaling pathway was the only shared term (Fig. 5A), indicating that Hippo signaling represents a central convergent node regulated by RIBOTAC. Based on this result, YAP1, the core transcriptional co-activator of the Hippo pathway, was selected as a candidate downstream effector for further investigation (Fig. 5, B and C). Kaplan-Meier survival analysis revealed that high YAP1 expression was associated with poorer overall survival in breast cancer patients (Fig. 5D). Functional assays demonstrated that YAP1 depletion significantly suppressed TNBC cell migration, as assessed by transwell and wound-healing assays (Fig. 5, E to M). The siRNA sequence of YAP1 was listed in table S2. At the molecular level, RIBOTAC treatment markedly reduced YAP1 protein level (Fig. 5, N and O, and fig. S11A). Proteomic KEGG analysis of RIBOTAC-regulated proteins showed significant enrichment in ubiquitin-related pathways (Fig. 5P). RIBOTAC treatment also induced a dose-dependent increase in global protein ubiquitination in TNBC cells (Fig. 5, Q and R, and fig. S11B). Notably, immunoprecipitation assays revealed that YAP1 ubiquitination was markedly enhanced following RIBOTAC treatment, indicating that RIBOTAC promotes YAP1 degradation through a ubiquitin-dependent pathway (Fig. 5S, and fig. S11C). CHX chase assays demonstrated that RIBOTAC treatment significantly shortened half-life of YAP1 in TNBC cells (Fig. 5, T and U, and fig. S11D). Moreover, co-treatment with chloroquine (CQ) did not rescue YAP1 protein levels, suggesting that RIBOTAC-induced YAP1 degradation is not mediated through autophagy (fig. S11, E and F). In contrast, the proteasome inhibitor MG-132 effectively rescued YAP1 protein levels despite RIBOTAC exposure (Fig. 5, V and W, and fig. S11G), demonstrating that RIBOTAC promotes YAP1 degradation via a proteasome-dependent mechanism. Importantly, pre-inhibition of hsa-microRNA-301a-3p significantly attenuated YAP1 degradation induced by RIBOTAC in MDA-BoM-1833 cells (Fig. 5, X and Y), suggesting that RIBOTAC-mediated YAP1 degradation is functionally dependent on hsa-microRNA-301a-3p.

Fig. 5. RIBOTAC targets Hippo signaling and promotes YAP1 ubiquitination.

(A) Venn diagram showing the intersection of downregulated KEGG pathways derived from mRNA-seq and miRNA-seq analyses following RIBOTAC treatment. (B) Western blot analysis of Hippo pathway-related proteins in MDA-MB-231 and MDA-BoM-1833 cells treated with RIBOTAC. (C) Quantification of YAP1 and p-YAP1 levels corresponding to (B). (D) YAP1 Kaplan–Meier survival analysis. (E) Western blot analysis confirming siRNA-mediated YAP1 knockdown in MDA-MB-231 and MDA-BoM-1833 cells. (F and G) Quantification of YAP1 protein levels corresponding to (E). (H) Transwell migration assay of TNBC cells following YAP1 knockdown. (I and J) Quantification of migrated TNBC cells from (H). (K) Wound-healing assay of MDA-MB-231 and MDA-BoM-1833 cells after YAP1 knockdown. (L and M) Quantification of migration ratio corresponding to (K). (N) Western blot analysis of YAP1 protein level after RIBOTAC treatment. (O) Quantification of YAP1 levels corresponding to (N). (P) KEGG enrichment analysis of RIBOTAC-regulated proteins. (Q) Western blot analysis of total ubiquitination levels in MDA-BoM-1833 cells following treatment with different concentrations of RIBOTAC. (R) Quantification of total ubiquitination corresponding to (Q). (S) Co-IP assay confirming RIBOTAC-enhanced YAP1 ubiquitination. (T) CHX chase assay assessing YAP1 protein stability after RIBOTAC exposure. (U) Quantification of YAP1 degradation corresponding to (T). (V) Western blot analysis of YAP1 protein following combined treatment with MG-132 and RIBOTAC. (W) Quantification of YAP1 protein expression corresponding to (V). (X) Western blot analysis of YAP1 protein levels in TNBC cells treated with RIBOTAC in the presence or absence of hsa-microRNA-301a-3p inhibition. (Y) Quantification of YAP1 protein expression corresponding to (X). Data are presented as mean ± SD (n = 3). Statistical significance was assessed using Student’s t test in (F), (G), (L), (M), and (R) and one-way ANOVA with Tukey’s multiple comparisons test in (C), (I), (J), (O), (U), (W), and (Y), where *P < 0.05, **P < 0.01, ***P < 0.001.

RIBOTAC restores NEDD4L expression through miRNA-mediated regulation

Given the proteasome-dependent degradation of YAP1 observed above, we next performed quantitative proteomic profiling in MDA-MB-231 cells to identify upstream regulators linking RIBOTAC treatment to YAP1 degradation. Volcano plot analysis revealed a subset of significantly upregulated and downregulated proteins following RIBOTAC treatment (Fig. 6A). To identify candidate genes regulated by hsa-microRNA-301a-3p, we integrated the proteomic data with miRNA target prediction analysis using TargetScan. qRT-PCR analysis demonstrated that RIBOTAC treatment robustly increased the mRNA levels of multiple predicted targets compared with control or Amikacin treatment, with NEDD4L showing the most pronounced induction (Fig. 6B). These proteomic and target-prediction analyses therefore linked RIBOTAC-mediated hsa-microRNA-301a-3p suppression to restoration of NEDD4L, providing a downstream mechanistic basis for YAP1 regulation. Consistent with its tumor-suppressive role, clinical dataset analyses revealed that NEDD4L expression was reduced in breast cancer and progressively decreased with advancing stage or subtype stratification (Fig. 6, C and D). Kaplan-Meier survival analysis further indicated that lower NEDD4L expression was associated with poorer overall survival in breast cancer patients (Fig. 6E). We next examined whether NEDD4L directly interacts with YAP1. Co-IP assays demonstrated that NEDD4L and YAP1 physically associated with each other (Fig. 6F). Immunofluorescence staining further confirmed cytoplasmic co-localization of NEDD4L and YAP1 in TNBC cells (Fig. 6, G and H).

Fig. 6. RIBOTAC restores NEDD4L expression.

(A) Volcano plot of proteomic profiling in MDA-MB-231 cells. (B) qRT-PCR analysis of predicted hsa-microRNA-301a-3p target genes in TNBC cells treated with control, Amikacin, or RIBOTAC. (C) Expression analysis of NEDD4L in breast cancer samples stratified by tumor stage. (D) Expression analysis of NEDD4L in breast cancer samples stratified by molecular subtype. (E) Kaplan-Meier survival analysis of breast cancer patients stratified by NEDD4L expression levels. (F) Co-IP assay showing endogenous interaction between NEDD4L and YAP1 in TNBC cells. (G) Immunofluorescence staining of NEDD4L and YAP1 in TNBC cells. (H) Line-scan analysis of fluorescence intensity profiles corresponding to (G). (I) Schematic illustration of luciferase reporter constructs containing wild-type or mutant NEDD4L 3’UTR. (J) Dual-luciferase reporter assay evaluating hsa-microRNA-301a-3p-mediated repression of NEDD4L 3’UTR activity. (K) Luciferase assay showing that RIBOTAC alleviates hsa-microRNA-301a-3p-mediated suppression of NEDD4L 3’UTR activity. (L and M) Actinomycin D chase assays evaluating NEDD4L transcript stability under the indicated conditions. (N) Western blot analysis of NEDD4L protein expression following hsa-microRNA-301a-3p inhibition in MDA-MB-231 cells. (O) Quantification of NEDD4L protein levels corresponding to (N). (P) Western blot analysis of NEDD4L expression in MDA-BoM-1833 cells treated with RIBOTAC or Amikacin. (Q) Quantification of NEDD4L protein levels corresponding to (P). (R) Immunofluorescence staining of NEDD4L in MDA-BoM-1833 cells following RIBOTAC treatment. (S) Quantitative analysis of NEDD4L fluorescence intensity corresponding to (R). Data are presented as mean ± SD (n = 3). Statistical significance was assessed using Student’s t test in (O) and (S) and one-way ANOVA with Tukey’s multiple comparisons test in (J), (K), and (Q), where *P < 0.05, **P < 0.01, ***P < 0.001.

To investigate whether the expression of NEDD4L is regulated by hsa-microRNA-301a-3p, we performed luciferase reporter assays using constructs containing either wild-type or mutant NEDD4L 3’UTR harboring the predicted hsa-microRNA-301a-3p binding site (Fig. 6I). Dual-luciferase assays showed that hsa-microRNA-301a-3p significantly suppressed luciferase activity of the wild-type NEDD4L 3’UTR reporter, whereas mutation of the binding site abolished this repression (Fig. 6J). Importantly, RIBOTAC treatment effectively relieved hsa-microRNA-301a-3p-mediated inhibition of NEDD4L 3’UTR activity (Fig. 6K). Consistently, actinomycin D chase assays showed that RIBOTAC treatment stabilized NEDD4L mRNA by delaying its decay, an effect comparable to that observed upon direct inhibition of hsa-microRNA-301a-3p (Fig. 6, L and M). Functionally, inhibition of hsa-microRNA-301a-3p increased NEDD4L protein expression (Fig. 6, N and O), and RIBOTAC treatment induced a dose-dependent upregulation of NEDD4L protein levels, whereas Amikacin alone produced only modest effects (Fig. 6, P and Q, and fig. S11H). Enhanced NEDD4L expression following RIBOTAC treatment was further confirmed by immunofluorescence staining (Fig. 6, R and S). These findings demonstrate that RIBOTAC restores NEDD4L expression by antagonizing hsa-microRNA-301a-3p-mediated repression, thereby providing a mechanistic link between RIBOTAC-triggered miRNA degradation and downstream YAP1 suppression.

DISCUSSION

Metastatic progression in TNBC arises from multilayered regulatory disruptions rather than aberrations in a single oncogenic protein. While extensive efforts have focused on inhibiting downstream signaling effectors, increasing evidence indicates that post-transcriptional regulatory networks play a central role in stabilizing invasive and therapy-resistant tumor states. MicroRNAs, as key coordinators of these networks, integrate signals across multiple pathways controlling migration, survival, and tissue colonization. However, despite their biological importance, oncogenic microRNAs have remained largely inaccessible to pharmacological intervention.

In this study, we identify hsa-microRNA-301a-3p as a clinically and functionally relevant regulator of TNBC metastasis. Its expression is markedly elevated in TNBC patient samples and is associated with poor clinical outcomes, consistent with previous reports implicating this microRNA in aggressive breast cancer behavior (13–15). Notably, integrative analysis of clinical datasets further revealed a strong association between hsa-microRNA-301a-3p expression and bone metastasis, suggesting a role in skeletal colonization. Functional validation in bone-metastatic mouse models confirmed that suppression of hsa-microRNA-301a-3p—either by genetic knockdown or pharmacological degradation—significantly reduced metastatic burden, alleviated osteolytic lesions, and limited tumor growth in vivo. These effects were accompanied by reduced migratory capacity in vitro, supporting a direct role for hsa-microRNA-301a-3p in metastatic dissemination.

Multiple strategies have been developed to inhibit microRNAs, including antisense oligonucleotides (ASOs), antagomirs, and related RNA-based strategies (34). Although these approaches show promise in preclinical models, their clinical translation is limited by delivery barriers, poor tissue distribution, and potential off-target effects (26–28). These challenges underscore the need for alternative pharmacological strategies to selectively target oncogenic microRNAs in solid tumors. Pharmacological degradation of microRNAs using RIBOTACs has established precedent. Early studies demonstrated the feasibility of this approach using Targaprimir-96-derived RIBOTACs, followed by dovitinib-reprogrammed RIBOTACs targeting pre-miR-21, and more recent designs directed against pre-miR-155 in cancer models (23, 27, 28). Compared with nucleic acid–based inhibitors, RIBOTACs harness small-molecule pharmacokinetics and RNase L–mediated catalytic degradation, enabling selective RNA cleavage while avoiding delivery barriers and hybridization-dependent off-target effects (26, 27). By linking the clinically approved aminoglycoside amikacin to an RNase L–recruiting moiety, we generated a bifunctional RIBOTAC capable of recognizing the Drosha cleavage region of pri-microRNA-301a and promoting RNase L-dependent degradation of the target RNA. hsa-microRNA-301a-3p represses the E3 ubiquitin ligase NEDD4L, thereby limiting physiological turnover of the Hippo pathway effector YAP1. Inhibition of this microRNA restores NEDD4L expression and reinstates endogenous YAP1 ubiquitination and degradation. In both cellular and animal models, RIBOTAC treatment selectively reduced hsa-microRNA-301a-3p levels, restored NEDD4L expression, and suppressed TNBC bone metastasis.

The Hippo effector YAP1 functions as a central regulator of organ size, stemness, and metastasis (35–38). Direct targeting of YAP1 has long been pursued through disruption of YAP-TEAD interactions, inhibition of TEAD activity, or genetic silencing strategies (39–47). While these approaches have demonstrated partial efficacy in preclinical models, their clinical translation has been hindered by limited selectivity, systemic toxicity, and adaptive signaling compensation (42). Recent studies have demonstrated that targeted degradation of endogenous YAP via nanobody-based bioPROTAC effectively suppresses tumor progression, reinforcing YAP as a therapeutically actionable vulnerability. However, such protein-centric strategies primarily eliminate YAP at the effector level and do not reestablish upstream regulatory control mechanisms (48). In contrast, restoring an endogenous regulatory circuit upstream of YAP1 offers a more context-dependent means of pathway modulation. By reinstating NEDD4L-mediated degradation, our strategy constrains YAP1 activity selectively in tumor cells characterized by aberrant hsa-microRNA-301a-3p expression, thereby reducing the risk associated with indiscriminate YAP1 suppression.

Although the NEDD4L-YAP1 axis represents a dominant downstream pathway mediating the effects of hsa-microRNA-301a-3p, this microRNA is unlikely to function through a single target alone. MicroRNAs typically exert pleiotropic effects, and additional targets of hsa-microRNA-301a-3p may contribute to extracellular matrix remodeling, immune modulation, or therapeutic resistance (49, 50). In this context, YAP1 serves as a key functional readout linking microRNA degradation to metastatic suppression, while broader regulatory consequences may further enhance therapeutic efficacy.

Beyond its biological implications, several challenges remain before clinical translation. Optimization of pharmacokinetic and biodistribution properties is required to ensure sustained target engagement and minimizing unintended RNA cleavage remains an important consideration. Furthermore, the limited structural understanding of RNA–small-molecule interactions remains a barrier to rational design. Advances in cryo–electron microscopy, AI-driven docking, and high-throughput RNA structure mapping are expected to accelerate the discovery of more potent and selective RNA degraders (51). Collectively, our study establishes hsa-microRNA-301a-3p as a pivotal oncogenic driver of TNBC bone metastasis and demonstrates that a rationally designed RIBOTAC can selectively degrade pri-microRNA-301a, restore the tumor-suppressive NEDD4L-YAP1 axis, and suppress metastatic progression both in vitro and in vivo. The selective and target-dependent activity of RIBOTAC is further supported by target-expression-dependent assays and functional competition with excess Amikacin (fig. S4), and the requirement for RNase L was validated by knockdown and rescue experiments (fig. S8). Moreover, NEDD4L perturbation experiments support that the anti-migratory effect of RIBOTAC is mediated, at least in large part, through the NEDD4L-YAP1 axis (fig. S12). These findings provide a proof-of-concept for the structure-guided development of RNA-degrading small molecules as a versatile therapeutic approach for targeting pathogenic non-coding RNAs in TNBC.

MATERIALS AND METHODS

Materials and reagents

Amikacin (TX20742) was purchased from Topscience (Shanghai, China). PTX (A4393) was purchased from APExBIO Technology LLC. All other chemical reagents and solvents were obtained from Med-ChemExpress, Adamas-beta, Sinopharm Chemical Reagent Co., Ltd., Energy Chemical, and CSBio (Shanghai).

Cell culture

MDA-MB-231, MCF10A, and HEK-293T cell lines were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The 4T1.2 cell line was obtained from HySigen Biosciences. The parental MDA-BoM-1833 cell line was kindly provided by Professor Yudong Zhou (University of Mississippi), and luciferase-labeled derivatives were generated in our laboratory. MDA-MB-231 cells were maintained in Leibovitz’s L-15 medium with 10% FBS and 1% penicillin–streptomycin (PS) at 37°C under CO₂-free conditions. MCF10A cells were maintained in specific medium (ZQ-1311, Zhongqiao Xinzhou Biotechnology) with 10% FBS and 1% PS. 4T1.2 cells were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium with 10% FBS and 1% PS. HEK-293T and MDA-BoM-1833 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) with the same supplements. Except for MDA-MB-231 cells, all other cells were incubated at 37°C with 5% CO₂.

RNA interference and lentiviral infection

All siRNAs and microRNA inhibitors were synthesized by GenePharma and Genomeditech (Shanghai, China), as listed in tables S1 and S2. RNase L (P63135) and NEDD4L (P100951) plasmids were purchased from Miaoling Biotechnology Co., Ltd. (Wuhan, China). The hsa-microRNA-301a-3p shRNA plasmid was purchased from Nanjing Jingmai Biotechnology (Nanjing, China), and the sequence (5’-GATCCGctttgacaatGATttgcactgCCGGGctttgacaatGATttgcactgTTTTTTG-3′) was cloned into the PE209 pLenti-shRNA NC-Luc2-Puro vector. Cells were seeded in 6-well plates at a density of 2 × 10⁵ cells per well and allowed to adhere for 24 h before transfection or infection. siRNAs were introduced using GP-transfect-mate (G04008, GenePharma, Shanghai, China) in accordance with the supplier’s protocol. MicroRNA inhibitors and plasmids were transfected with Lipofectamine 2000 Reagent (11668019, Invitrogen, USA) according to routine transfection procedures. For lentiviral experiments, shRNA plasmids were packaged into 293T cells using ExFect Transfection Reagent (T101–01, Vazyme), and the resulting viral supernatants were subsequently used to infect target cells following Genomeditech’s guidelines. Transfection or infection was terminated after 48 h, and cells were harvested for subsequent analyses.

RNA extraction and qRT-PCR

Cellular RNA was extracted with the FastPure Cell/Tissue Total RNA Isolation Kit V2 (RC112, Vazyme) following the supplier’s protocol. To analyze mRNA and pri-microRNA expression, 1 μg of total RNA was converted into cDNA using HiScript II Q RT SuperMix for qPCR (R223, Vazyme), and qPCR reactions were set up with ChamQ Universal SYBR qPCR Master Mix (Q711, Vazyme) on a LightCycler 96 system (Roche, Switzerland). MicroRNA expression was examined using a stem-loop reverse transcription strategy with the miRNA 1st Strand cDNA Synthesis Kit (MR101, Vazyme), followed by amplification using miRNA Universal SYBR qPCR Master Mix (MQ101, Vazyme). GAPDH was selected for normalization of mRNA and pri-microRNA levels, while U6 was used for microRNA normalization. Relative transcript abundance was determined by the 2^−ΔΔCt method, and primer information is listed in table S3.

Transwell migration assay

Cell migration assay was performed using 8 μm pore size transwell chambers (3422, Corning), as previously described with minor modifications (52). For cells transfected with vector, siRNA, or microRNA inhibitor, 2 × 10⁴ cells per 100 μL basic medium (without FBS) were seeded into the upper chamber and incubated for 24 hours. The lower chambers were filled with 600 μL complete medium containing FBS to serve as a chemoattractant. Following a 24-hour migration period, cells on the lower surface of the polycarbonate membrane were immobilized with 4% paraformaldehyde and visualized by crystal violet staining. Images were then acquired and analyzed with ImageJ software.

Real-time cell analysis (RTCA) assay

Cell proliferation and migratory activity were dynamically evaluated with the xCELLigence RTCA platform (Agilent, USA). This system records changes in electrical impedance and converts them into cell index values. For migration detection, 1.5 × 10⁴ transfected cells, including vector-, siRNA-, or microRNA inhibitor-treated cells, were resuspended in basic medium and placed in the upper compartment of CIM-Plates. Complete medium supplemented with 10% FBS was added to the lower compartment to establish a chemotactic gradient. The cell index was automatically recorded over a 36-hour period. In drug-intervention experiments, cells were seeded in the upper compartment and treated with Amikacin or RIBOTAC at the indicated concentrations. Migration-associated impedance changes were continuously monitored for 36 hours and analyzed with the RTCA software.

Scratch assay

Cells (3.5 × 10⁵ cells/well) were seeded in 12-well plates and cultured overnight to form a confluent monolayer. When the confluence reached approximately 95%, a straight scratch was generated across the center of each well using a sterile pipette tip. Baseline images were captured using Cytation 5 microplate imaging system (BioTek, USA). Cells were then incubated with the indicated compounds for 24 hours, after which the scratched areas were imaged again at the same positions. The migration rate was calculated using the formula

$$\begin{array}{c}\text{Migration rate} (\%)=\text{Scratch healing rate} (\%)\hfill \\ =\left[(\text{Scratch area at 0 hours}-\text{Scratch area at 24 hours})/\right.\hfill \\ \left.\text{Scratch area at 0 hours}\right]\times 100\%\hfill \end{array}$$

Colony formation assay

Long-term proliferative potential was determined using a colony formation assay. Briefly, transfected or compound-treated cells were seeded into 12-well plates at 5 × 10² cells per well and maintained for 10–14 days to allow colony development. The resulting colonies were then fixed with 4% paraformaldehyde, visualized by 0.5% crystal violet staining, and imaged with a Cytation 5 system (BioTek, USA).

Molecular docking

The primary sequence of hsa-microRNA-301a-3p was obtained from miRBase, and its 3D structure was predicted using MC-Fold/MC-Sym. The chemical structures of candidate small molecules were downloaded from the PubChem database and energy-minimized. Molecular docking was performed using MOE 2022 (Chemical Computing Group, Montreal, Canada). The docking site was defined as the Drosha cleavage region of pri-microRNA-301a, and the top-scoring binding poses were selected based on docking scores and predicted interaction patterns.

Microscale thermophoresis (MST) binding measurements

MST fluorescent measurements were performed following the method described previously (27). The measurements were conducted on a Monolith NT.115 system (NanoTemper, Germany) using the fluorescence of a hsa-microRNA-301a-3p hairpin or base paired control hairpin purchased from Sangon Biotech with RNase-free HPLC purification. The sequences and structural features of the RNA oligonucleotides used in MST assays are summarized in table S4. The RNA (5 nM) was diluted in 1× MST Buffer (8 mM Na₂HPO₄, 190 mM NaCl, 1 mM EDTA, and 0.05% (v/v) Tween-20), folded by heating at 60°C for 5 min, and slowly cooled to room temperature. Compounds were diluted in 1× MST Buffer, followed by 1:3 dilutions in 1× MST Buffer containing 5 nM RNA. Samples were incubated for 30 min at room temperature, loaded into the Monolith NT.115 capillary (MO-K022, NanoTemper, Germany), and analyzed with Monolith NT.115 system.

In vitro fluorescent RNA cleavage

A hsa-microRNA-301a-3p Hairpin Precursor RNA labeled with a 5′ Fluorescein (FAM) and 3’Black Hole Quencher (BHQ-1) (5′ FAM-ACUGCUAACGAAUGCUCUGACUUUAUUGCACUACUGUACUUUACAGCUAGCAGUGCAAUAGUAUUGUCAAAGCAUCUGAAAGCAGG–3’ BHQ-1) was purchased from Sangon Biotech with HPLC purification. RNA (100 nM) was folded in 1× RNase L Buffer by heating at 65°C for 5 min followed by slow cooling to room temperature (25 mM Tris-HCl, pH 7.4, 100 mM KCl) without MgCl₂, β-mercaptoethanol or ATP. After folding, reactions were supplemented with 10 mM MgCl₂, fresh 7 mM β-mercaptoethanol, and 50 μM ATP. Next, 100 nM of RNase L, prepared as described previously (27), was incubated with RNA and candidate compounds in 1× RNase L Buffer. The reaction mixtures were loaded into white opaque 96-well plates and maintained at room temperature for 60 min. Fluorescence signals were then detected at Ex/Em wavelengths of 485/525 nm using a Cytation 5 microplate reader (BioTek, USA).

Western blotting analysis

Cells from different treatment groups were harvested and subjected to protein extraction using NP40 lysis buffer. The total protein content was measured with a BCA assay kit (P0010S, Beyotime). Equivalent amounts of protein were fractionated by SDS-PAGE, followed by transfer onto PVDF membranes. Then membranes were blocked with 5% BSA and incubated overnight at 4°C with primary antibodies, including NEDD4L (13690–1-AP, Proteintech), YAP1 (A21216, ABclonal), p-YAP1 (AP1398, ABclonal), Ubiquitin (ET1609–21, HuaAn Biotechnology), and GAPDH (LF205, Epizyme Biotech). After washing, the membranes were incubated at room temperature for 1 hour with HRP-conjugated anti-rabbit or anti-mouse IgG secondary antibodies (LF101, LF102, Epizyme Biotech). Protein bands were detected using a ChemiDoc MP imaging system (Bio-Rad, USA), and band intensities were quantified with ImageJ software.

Co-immunoprecipitation (Co-IP) for LC-MS/MS analysis

For Co-IP coupled with LC-MS/MS, cells cultured in 100 mm dishes were treated with RIBOTAC at 10 μM for 48 hours. Cell lysates were prepared in NP40 buffer (P0013F, Beyotime) supplemented with protease/phosphatase inhibitors. A total of 500 μg protein lysate was mixed with 2 μg NEDD4L antibody (13690–1-AP, Proteintech) and gently rotated overnight at 4°C. Immune complexes were pulled down using Protein A/G PLUS-Agarose beads (Santa Cruz, sc-2003) for 4 hours at 4°C. The bead-bound complexes were recovered by centrifugation, washed three times with NP40 buffer, and sent to OE Biotech Co., Ltd. (Shanghai, China) for LC-MS/MS-based proteomic profiling. Proteins with P < 0.05 and FC > 1.2 were considered differentially enriched.

Co-IP assays

Cells were seeded in 100-mm dishes and allowed to adhere overnight, followed by treatment with RIBOTAC (10 μM) for 48 hours. Cells were harvested and lysed in NP40 lysis buffer. Equal amounts of total protein (500 μg) were incubated with 2 μg of anti-YAP1 antibody (A21216, ABclonal) at 4°C overnight with gentle rotation. Protein A/G PLUS-agarose beads were then added, and the mixtures were further incubated for 4 hours at 4°C. The immunocomplexes were collected by centrifugation and washed three times with NP40 lysis buffer. The beads were subsequently resuspended in 3× loading buffer and boiled for 10 min. Immunoprecipitated proteins were analyzed by Western blotting using antibodies against YAP1, ubiquitin and GAPDH.

Immunofluorescence

Cells (4×10⁴ cells/ dish) were grown in 20 mm confocal dishes (BS20GJM, BioSharp) overnight and treated with RIBOTAC (10 μM) for 48 hours. Cells were rinsed with PBS, fixed with paraformaldehyde (P0099, Beyotime), and then permeabilized with 1% Triton X-100 (ST795, Beyotime). After blocking with 1% BSA for 1 hours at room temperatureCells were incubated with primary antibody overnight at 4°C, followed by fluorophore-conjugated secondary antibodies for 2 hours. Nuclei were visualized with DAPI staining for 10 min, and images were obtained on a ZEISS LSM 710 single-photon laser scanning confocal microscope (Carl Zeiss, Germany).

Dual luciferase assay

HEK-293T Cells were seeded into 48-well plates and cultured to approximately 40% confluency. Cells were then transiently co-transfected with a Firefly luciferase plasmid encoding the 3’ UTR of NEDD4L and a control Renilla luciferase plasmid using ExFect Transfection Reagent (T101–01, Vazyme). The sequences of microRNA mimics and NEDD4L 3’ UTR reporter constructs used in dual-luciferase assays are summarized in table S5. After 24 hours, the growth medium was replaced with fresh medium containing the indicated compounds. Cells were incubated for an additional 48 hours, followed by lysis and measurement of Firefly and Renilla luciferase activities using the Dual-Luciferase Reporter Assay Kit (RG027, Beyotime). The luminescence signals were detected using Cytation 5 at 560 nm and 465 nm, respectively.

YAP1 degradation experiments

Cells were seeded into 6-well plates at a density of 2.0 × 10⁵ cells per well and allowed to adhere overnight. For dose–response assays, cells were treated with RIBOTAC (0, 1, 3, 10, or 30 μM) for 48 hours. For time-course assays, cells were incubated with 10 μM RIBOTAC for 0, 12, 24, 36, or 48 hours. For mechanistic rescue assays, cells were treated with 10 μM RIBOTAC for 48 hours, with additional supplementation of 10 μM MG-132 or 10 μM chloroquine (CQ) during the final 24 hours, or 10 μg/mL cycloheximide (CHX) for 12, 24, or 48 hours, as indicated. After treatment, cells were collected for Western blot analysis of YAP1 to evaluate protein stability and degradation.

RNA sequencing analysis

Exponentially growing MCF10A, MDA-MB-231, MDA-BoM-1833, and MDA-MB-231 cells treated with 10 μM RIBOTAC for 48 hours, were harvested at a density of approximately 1 × 10⁶ cells per sample. Total RNA was extracted using TRIzol reagent (B511311, Sangon Biotech, Shanghai, China) according to the manufacturer’s instructions. Both mRNA sequencing (mRNA-seq) and microRNA sequencing (microRNA-seq) were performed and analyzed by OE Biotech Co., Ltd. (Shanghai, China). The criteria for screening differentially expressed genes (DEGs) between treatment groups were P < 0.05 and fold FC > 2.

Quantitative proteomics analysis

MDA-MB-231 cells were exposed to PBS or 10 μM RIBOTAC for 48 hours, with approximately 1 × 10⁶ cells prepared for each sample. Cells were lysed on ice for 30 min using NP-40 buffer supplemented with phosphatase and protease inhibitor cocktails. Total protein concentrations were determined with a BCA protein assay kit (P0010S, Beyotime). Proteins were precipitated with acetone overnight, collected by centrifugation, and resuspended in 8 M urea. After reduction with 0.1 M dithiothreitol (DTT) and alkylation with 0.5 M 2-iodoacetamide (IAA), samples were digested with trypsin overnight at 37°C under agitation. Then peptides were desalted using SPE spin columns, followed by lyophilization as previously described (53–54). LC-MS/MS analysis was performed on a Q Exactive Plus Orbitrap mass spectrometer coupled to an EASY-nLC system (Thermo Fisher Scientific, USA). Peptides were separated on a 25 cm C18 analytical column using a water–acetonitrile gradient containing 0.1% formic acid. MS1 spectra were acquired in the Orbitrap at a resolution of 120,000, whereas MS2 spectra were generated by CID fragmentation in the ion trap. MS3 scans were obtained after HCD activation with synchronous precursor selection and detected in the Orbitrap at a resolution of 60,000 or 50,000.

Raw MS files were processed using Proteome Discoverer 2.4 against the Homo sapiens UniProt database. Protein and peptide identifications were filtered at a false discovery rate (FDR) of 1%. Quantitative values were obtained by label-free quantification (LFQ) and normalized using MaxLFQ, with a minimum ratio count of 2. Statistical comparisons were performed using an unpaired two-sided t test. Statistical analysis was based on an unpaired two-sided t test, followed by Benjamini–Hochberg correction. Differentially expressed proteins (DEPs) between the two groups were defined as those exhibiting P < 0.05 and FC > 1.5.

Animal experiment

Bone metastasis models were established according to a previously reported method (54). Female BALB/c nude mice, 4–6 weeks of age and weighing 18–20 g, were obtained from Shanghai Slack Laboratory Animal Co., Ltd. Animals were maintained under specific pathogen-free conditions. The animal study protocol was approved by the Ethics Committee of Shanghai University of Traditional Chinese Medicine (Approval No.: PZSHUTCM2501170007). All animal experiments were performed in accordance with the ARRIVE guidelines.

Bone metastasis models were established by intracardiac injection of 1 × 10⁵ MDA-BoM-1833-luc cells or 1.5 × 10⁴ 4T1.2-luc cells into the left ventricle. Injection accuracy was verified using cardiac ultrasound, after which mice were randomized into treatment groups receiving PBS, RIBOTAC at 10 or 25 mg/kg, PTX at 5 mg/kg, or zoledronic acid at 50 μg/kg. Metastatic burden was monitored by IVIS imaging following intraperitoneal injection of 20 mg/kg D-luciferin. At the experimental endpoint, bone morphology was assessed using micro-computed tomography (micro-CT; Quantum GX, PerkinElmer, USA). Whole blood, major organs, and bone tissues were collected for subsequent analyses. All histological processing and staining, including H&E and FISH, were performed by Wuhan Servicebio Technology Co., Ltd. (Wuhan, China). The sequence of the FISH probe for hsa-microRNA-301a-3p was 5’-GCTTTGACAATACTATTGCACTG-3′.

Hemolysis assay

The hemolytic potential of RIBOTAC was assessed using mouse erythrocytes. Briefly, erythrocytes were isolated from fresh mouse blood by centrifugation at 1000g for 30 min at 4°C, washed three times with cold 0.9% NaCl, and seeded into 96-well plates. Cells were treated with RIBOTAC or 0.1% Triton X-100 for 1 hour at 37°C, followed by centrifugation at 1000g for 10 min. Hemoglobin released into the supernatant was determined by measuring absorbance at 570 nm with a Cytation 5 reader (BioTek, USA). PBS-treated erythrocytes were used as the negative control, and 0.1% Triton X-100-treated erythrocytes represented complete hemolysis. Hemolysis was calculated as: (A_RIBOTAC − A_PBS)/ (A_PC − A_PBS) × 100%.

Enzyme-linked immunosorbent assay (ELISA)

Blood samples collected from mice in the bone metastasis tumor model were centrifuged at 3000 rpm for 20 min at 4°C, and the resulting supernatants were used for ELISA assays. The serum levels of bone-resorption marker β-CTX and bone-turnover-associated marker BSP were quantified using ELISA kits (BPE20586, BPE20831, Lengton Bioscience, Shanghai, China) according to the manufacturer’s instructions.

Statistical analysis

Statistical analysis was performed with Student’s t test or ANOVA with Tukey’s multiple comparison tests via GraphPad Prism 10.3.1. All data are presented as means ± SD. In vitro experiments were performed at least three independent times unless otherwise stated, and animal data represent biological replicates from individual mice. Statistical significance levels were defined as *P < 0.05, **P < 0.01, ***P < 0.001. ns. indicates no significant difference.

Supplementary Materials

This PDF file includes:

Figs. S1 to S12

Tables S1 to S5

References